AI-Assisted Development Tools
Explore how AI-assisted development tools can streamline AWS application coding by generating specs, suggesting code, and automating reviews. Learn to use tools like Kiro for spec-driven code generation, code completion, refactoring, and security scanning. Understand the importance of reviewing AI-generated code carefully, verifying IAM permissions, and maintaining secure, tested, and optimized cloud applications.
Writing an AWS application involves a lot of routine code: SDK calls, IAM policies, infrastructure definitions, tests, and glue between services. AI-assisted development tools take on part of that work. They suggest code as we type, turn a feature description into a plan and then into code, propose refactors, and flag problems in code we have already written.
However, an AI-generated Lambda function can run and still use an overly broad IAM policy or a hardcoded secret. The skill the exam tests is using the tools to generate, review, and optimize code while checking the output ourselves. This lesson covers the main kinds of assistance and the review habits that go with them.
The diagram below shows where each kind of assistance fits in the development loop.
Spec-driven code generation with Kiro
Kiro is an AWS agentic development tool, available as an IDE and a command-line interface (CLI), that writes code from a description of what we want. It supports two ways of working. In a vibe session, we describe a change in conversation, and Kiro makes it. In a spec session, Kiro first turns the request into a written specification and only then writes code. Spec-driven generation suits larger features because the plan is written down and can be corrected before any code exists.
A Kiro spec is a set of three Markdown files, created in this order:
requirements.md: User stories with acceptance criteria written in EARS notation ...